MareNostrum 5, the eighth most powerful supercomputer
MareNostrum 5, the supercomputer of the Barcelona Supercomputing Center, is one of the most powerful in the world. With technology from Eviden (Atos), it ranks eighth in the TOP500.
The BSC is the only European center with two entries in the top 20, the other is in 19th place, showcasing its prowess in both general-purpose and accelerated computing.
The unique architecture combines a general-purpose partition for classical computing with an accelerated partition focusing on artificial intelligence, powered by NVIDIA Hopper processors. MareNostrum 5 features substantial storage capacity and an InfiniBand NDR200-based interconnection network.
The characteristics of BSC supercomputers are as follows:
MareNostrum 5 ACC (eighth place)
The machine has 6408 nodes based in Intel Sapphire rapids. Each is configures as:
- 2x Intel Sapphire Rapids 8480+ at 2Ghz and 56c each (112 cores per node)
- 256 GB of Main memory, using DDR5 (with 216 nodes with 1024GB)
- 4x Nvidia H100 with a total 4.500 GPUs
- 960GB on NVMe storage (/scratch)
- 1x NDR200 shared by 2 nodes (SharedIO) (BW per node 100Gb/s)
MareNostrum 5 GPP (nineteenth place)
The machine has 1120 nodes based in Intel Sapphire rapids and Nvidia Hopper GPUs. Each is configures as:
- 2x Intel Sapphire Rapids 8460Y+ at 2.3Ghz and 32c each (64 cores node)
- 512 GB of Main memory, using DDR5
- 4x Nvidia Hopper GPUs with 64 HBM2 memory
- 460GB on NVMe storage (/scratch)
- 4x NDR200 (BW per node 800Gb/s)
Recognized as the “Top Supercomputing Achievement of 2023” by HPCWire readers, it is praised for integrating traditional supercomputing with AI for groundbreaking scientific exploration. The commissioning of MareNostrum 5 will reinforce BSC’s position as a leading European supercomputing center.
TOP 10 System – November 2023
| Rank | System | Cores | Rmax (PFlop/s) | Rpeak (PFlop/s) | Power (kW) |
|---|---|---|---|---|---|
| 1 | Frontier – HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11, HPE DOE/SC/Oak Ridge National Laboratory United States | 8,699,904 | 1,194.00 | 1,679.82 | 22,703 |
| 2 | Aurora – HPE Cray EX – Intel Exascale Compute Blade, Xeon CPU Max 9470 52C 2.4GHz, Intel Data Center GPU Max, Slingshot-11, Intel DOE/SC/Argonne National Laboratory United States | 4,742,808 | 585.34 | 1,059.33 | 24,687 |
| 3 | Eagle – Microsoft NDv5, Xeon Platinum 8480C 48C 2GHz, NVIDIA H100, NVIDIA Infiniband NDR, Microsoft Microsoft Azure United States | 1,123,200 | 561.20 | 846.84 | |
| 4 | Supercomputer Fugaku – Supercomputer Fugaku, A64FX 48C 2.2GHz, Tofu interconnect D, Fujitsu RIKEN Center for Computational Science Japan | 7,630,848 | 442.01 | 537.21 | 29,899 |
| 5 | LUMI – HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11, HPE EuroHPC/CSC Finland | 2,752,704 | 379.70 | 531.51 | 7,107 |
| 6 | Leonardo – BullSequana XH2000, Xeon Platinum 8358 32C 2.6GHz, NVIDIA A100 SXM4 64 GB, Quad-rail NVIDIA HDR100 Infiniband, EVIDEN EuroHPC/CINECA Italy | 1,824,768 | 238.70 | 304.47 | 7,404 |
| 7 | Summit – IBM Power System AC922, IBM POWER9 22C 3.07GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM DOE/SC/Oak Ridge National Laboratory United States | 2,414,592 | 148.60 | 200.79 | 10,096 |
| 8 | MareNostrum 5 ACC – BullSequana XH3000, Xeon Platinum 8460Y+ 40C 2.3GHz, NVIDIA H100 64GB, Infiniband NDR200,EVIDEN EuroHPC/BSC Spain | 680,960 | 138.20 | 265.57 | 2,560 |
| 9 | Eos NVIDIA DGX SuperPOD – NVIDIA DGX H100, Xeon Platinum 8480C 56C 3.8GHz, NVIDIA H100, Infiniband NDR400, Nvidia NVIDIA Corporation United States | 485,888 | 121.40 | 188.65 | |
| 10 | Sierra – IBM Power System AC922, IBM POWER9 22C 3.1GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM / NVIDIA / Mellanox DOE/NNSA/LLNL United States | 1,572,480 | 94.64 | 125.71 | 7,438 |